【摘要】 基于相空间重构思想,采用最大Laypunov指数法确定洞庭湖区岳阳水文站30 a的月降雨序列具有明显的混沌特性。应用混沌径向基函数神经网络预测月降雨量,预测精度远远低于时间序列分解模型,并表现出高度的无规律性。定量分析了噪声对混沌预测精度的影响,表明监测误差是影响混沌预测精度的一个重要因素,提高数据精度是提高混沌预测精度的一个有效方法。更多还原

【Abstract】 Based on phase-space reconstruction theory,embedding theory and maximum Lyapunov exponent,the chaotic behavior of monthly rainfall of Dongting Lake was recognized.Then chaos radial basic function neural network(CRBFNN) was applied to predict rainfall.Compared with the predictions of Time Series Decomposing Model(TSDM),the prediction of CRBFNN was much worse than the predictions of TSDM and showed non-regularity.At last,affect of noise to prediction precision was analyzed and the results indicated that monitoring error was an important source of ill-prediction of CRBFNN.更多还原